Login required to started new threads

Login required to post replies

Real time CdA in Zwift?
Quote | Reply
I think it would be fascinating to have a real time CdA display in Zwift. You'd need to access gradient, speed and power. Gradient and power should be accessible directly from the trainer or other PM, is there any way to access real-time virtual speed from the game? You'd assume a reasonable crr and weight for the bike and know your own weight.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
Not real-time, but post hoc you can use Aerolab in GC. There's no wind, but there's drafting and Crr changes from changing road surface.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
Would be cool to use Xbox kinect camera or similar in real time to calculate balpark cda. At least to know if you are hoods, TT, drops
Quote Reply
Re: Real time CdA in Zwift? [MadTownTRI] [ In reply to ]
Quote | Reply
I had considered some screen capture and character recognition to get this but seems very crude and I think I'd need better speed resolution.

Aerolab option is a good idea.

In these covid times my main fix is a weekly TTT on Zwift. My team has really been getting into it and we've improved from bottom 20% of the B (WTRL Frappe) league to the 60-70% region. Last week we were at about 58% which is a huge leap for us. Long term goal is to consistently get into the top 50%. I was hoping that a tool like this would help us position more efficiently, besides being a fun thing to do.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
carlosflanders wrote:
I had considered some screen capture and character recognition to get this but seems very crude and I think I'd need better speed resolution.

Aerolab option is a good idea.

In these covid times my main fix is a weekly TTT on Zwift. My team has really been getting into it and we've improved from bottom 20% of the B (WTRL Frappe) league to the 60-70% region. Last week we were at about 58% which is a huge leap for us. Long term goal is to consistently get into the top 50%. I was hoping that a tool like this would help us position more efficiently, besides being a fun thing to do.


In 2019 I helped a world tour team prep for the Vuelta TTT. They ended up winning it for the first time ever. They did 5 full days of TTT testing on the road, plus numerous hours testing physiology + lots of race history data.

We did a few things. We aero tested every rider. We analyzed trial runs to measure the efficiency of the rider in position 2, 3,.....We also modeled W'/CP to figure out depletion of the rider. We did multiple tests to confirm all the data

We then created a plan that took this all into consideration and paced it to have every rider empty at the end. We considered terrain, weight, aero, efficiency to get "back on the train"....many many things. Psychology of the riders included.

With previous rides you can figure out some of this. With more testing you can figure out the rest and optimize your next TTT. At the end you optimize your pull times.

It's fun !
Quote Reply
Re: Real time CdA in Zwift? [marcag] [ In reply to ]
Quote | Reply
The camera would be cool. But people would cheat that also.

Real life the picture at local TT’s makes a LOT more sense looking at the podium in any category than Zwift does. Sorry, I haven’t seen a Clyde in the tops of most any cat of a Usac TT. Zwift D, C, and even B, oh yeah you see it.

It’s a videogame, but still.
Quote Reply
Re: Real time CdA in Zwift? [burnthesheep] [ In reply to ]
Quote | Reply
I think some may have misinterpreted what I meant. I was talking about a display of the CdA that Zwift uses and changes dynamically depending on draft. Camera would read speed off the screen display, not capture rider position.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
In theory it's really easy, using the Developer API, or one of the APIs based on the Developer API. I'll give it a try in the next day or two. I mean easy to get your speed, and some other data. Not combining it all with data from other sources.
Last edited by: trail: Sep 2, 20 20:38
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
RGT already does it, kind of. They don't show your CDA in real time, but they do show you how many watts you are saving in the draft, and that changes in real time, depending on your position in the draft. That's basically CDA
Quote Reply
Re: Real time CdA in Zwift? [MRid] [ In reply to ]
Quote | Reply
MRid wrote:
RGT already does it, kind of. They don't show your CDA in real time, but they do show you how many watts you are saving in the draft, and that changes in real time, depending on your position in the draft. That's basically CDA
RGT?
Quote Reply
Re: Real time CdA in Zwift? [MRid] [ In reply to ]
Quote | Reply
Thanks. Didn't know this. Yes, it's the same info.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
CdA in Zwift is calculated based on rider height, bike and wheels meaning it is a fixed value that doesn't change as you ride. (https://zwiftinsider.com/...ects-speed-in-zwift/). Drafting and/or powerups modify the speed to watts calculation but its not a dynamic CdA calculations.

I would guess Zwift has intentionally avoided providing the exact CdA formula as that would only help people game the system and probably raise a ton of questions surrounding other elements of realism in the game. Just looking at speed to watts pattern my feeling is that Zwift low balls CdA and drastically over calculates the speed penalty of climbing. This keeps average speeds up and helps groups stay together on the flats while enhancing separation on the climbs. This is not a criticism as I actually think its works extremely well, but it just isn't very real world accurate. If they published their formulas I think they would be taken to task by people demanding a 'realistic' racing situation.
Quote Reply
Re: Real time CdA in Zwift? [Ai_1] [ In reply to ]
Quote | Reply
RGT (formally Road Grand Tours, I think), it's a zwift competitor. In my mind, it's the closest to being a real competitor to Zwift, though it's not there yet. I know some people like Rouvy, but it just didn't work for me. Video is something that sounds good in theory, but in practice it actually takes away from the immersion. RGT is rendered, similiar to zwift.

A couple of us on the DIRT team have been playing with RGT, and it has some potential. I think the draft and pack dynamics are simulated better on RGT than zwift. RGT actually computes everyone's position, both North/South and East/West on the 2D plan, as opposed to just North/South on Zwift. As a result, each rider has it's own computed draft effect, base on positioning, rather than the group draft algorithm that Zwift seems to use. As a result, in my opinion (and others of us that have been playing with it), RGT seems to provide a more realistic draft/pack effect.

The graphics on RGT are pretty decent, similiar to Zwift. I'd say Zwift is probably more polished. Neither Zwift, nor RGT, have a very intuitive UI, I don't know why this is so hard for Cycling apps, but it appears to be.

There social aspect of RGT is just not going to compare to Zwift at all. They don't have the user base or the infrastructure set up to even really compete at this point. (no zwiftpower for RGT, the group ride community is pretty much nonexistant). There are some Races on RGT, but there aren't as many as Zwift (not even close). And there are no categories, everyone just races together. RGT does not allow zPower, you have to have a power source, so that helps with some of the issues in Zwift racing, but it's still online racing, so weight doping, dodgy powermeters, etc. are still going to be an issue.

https://www.youtube.com/watch?v=TG5rnWVftXM Here's a video that compares racing on the 2 platforms. I think it does a pretty good job of comparing.


Probably, the best thing RGT has going for it, is it's magic roads feature. Basically you upload a GPX file, and it creates the course for you, and then you can ride it (or race it). Using the strava route creator, you can essentially create a course from anywhere you want. It works pretty well. There are a few kinks, especially if the gpx file has sudden gradient changes or really sharp turns. But overall it works pretty well. This is the feature I use the most.

RGT is free to use, though you need to be premium to use all the features (like magic roads). Premium is $10 a month, but it's free during Covid, which I think they are still doing.

Anyway, in my mind it's not a Zwift replacement at this point, but its got the most promise of any of the ones I've seen. I enjoy riding on it.
Quote Reply
Re: Real time CdA in Zwift? [MadTownTRI] [ In reply to ]
Quote | Reply
MadTownTRI wrote:
Would be cool to use Xbox kinect camera

I've had a few calls in the past by a company that was doing something like this.
Can't remember the name, maybe Stac virtual wind tunnel?

Brian Stover USAT LII
Accelerate3 Coaching
Insta

Quote Reply
Re: Real time CdA in Zwift? [MRid] [ In reply to ]
Quote | Reply
Thanks, I've heard of Road Grand Tours but didn't connect the acronym.
Quote Reply
Re: Real time CdA in Zwift? [MadTownTRI] [ In reply to ]
Quote | Reply
MadTownTRI wrote:
Would be cool to use Xbox kinect camera or similar in real time to calculate balpark cda. At least to know if you are hoods, TT, drops

I think it would be great if Zwift could be played on Xbox or Playstation. I'd rather get a purposeful gaming unit than a gaming PC but sadly this is a no go.
Quote Reply
Re: Real time CdA in Zwift? [scott8888] [ In reply to ]
Quote | Reply
scott8888 wrote:
CdA in Zwift is calculated based on rider height, bike and wheels meaning it is a fixed value that doesn't change as you ride. (https://zwiftinsider.com/...ects-speed-in-zwift/). Drafting and/or powerups modify the speed to watts calculation but its not a dynamic CdA calculations.

I would guess Zwift has intentionally avoided providing the exact CdA formula as that would only help people game the system and probably raise a ton of questions surrounding other elements of realism in the game. Just looking at speed to watts pattern my feeling is that Zwift low balls CdA and drastically over calculates the speed penalty of climbing. This keeps average speeds up and helps groups stay together on the flats while enhancing separation on the climbs. This is not a criticism as I actually think its works extremely well, but it just isn't very real world accurate. If they published their formulas I think they would be taken to task by people demanding a 'realistic' racing situation.
Do you think weight is NOT factored into the aerodynamic drag calculation?
I assumed it was but I don't know.
Quote Reply
Re: Real time CdA in Zwift? [carlosflanders] [ In reply to ]
Quote | Reply
My problem would be all the "likes" sticking out of my back pockets.
Quote Reply
Re: Real time CdA in Zwift? [Ai_1] [ In reply to ]
Quote | Reply
I don't know for sure but I don't think weight is taken into account with aerodynamics. Once you account for height, in the way they have done, I imagine the relationship between front area and weight is very very weak.

Frontal area is largely a function of a persons width whereas weight is largely a function of depth. You can get a very good guess on a persons width based on their height but weight doesn't follow the same proportionality (at least not in the US of UK) although maybe in theory it should.

Outside the very small number of people who seriously race on Zwift I also think it would be counter productive for Zwift to use weight to work out aerodynamics. Zwift should be social and encouraging and the weight penalty associated with climbing is already very large. I don't see how double penalizing people carrying a few extra pounds would be a benefit. Plus is would further incentivize weight doping. My experience is people believe they are taller and lighter than they are.
Quote Reply
Re: Real time CdA in Zwift? [scott8888] [ In reply to ]
Quote | Reply
If only there were a way to see whether CdA varied with weight.
Quote Reply
Re: Real time CdA in Zwift? [desert dude] [ In reply to ]
Quote | Reply
We did that https://www.velogicfit.com/3d-aero
Stac is more about scanning and post modelling
Quote Reply
Re: Real time CdA in Zwift? [RChung] [ In reply to ]
Quote | Reply
I’m curious about what you are suggesting because I have a professional interest in big data and the issues surrounding bias introduced when doing these sorts of calculations.

Zwift knows rider weight, height and sex. You can’t directly calculate cda from these values but you can develop a model/equation to estimate it. The issue is that these are not fully independent variables and will entail a lot of cross correlations. You have to for example normalize weight for height, sex and age when isolating weight as a predictor of cda. How this is done introduces bias into the model by favoring some variables over others.

Assuming you do have the required large database containing measured cda along with rider information what rider information would you choose to start with? Zwift’s choice of height seems right to me and I suspect once you normalize the data for height you are left with a lot of noise. But I bet you could also ignore height and use sex, weight and age to describe cda and you would reach a slightly different answer. Sex and age may actually be sufficient in this case because I’ll bet any database of measured cda is strongly biased to pro athlete physiology. Which brings up another issue surrounding the representativeness of your population to the data used to model your population.

Alternatively you could take a few riders measure their cda, fatten then up and remeasure cda. Then develop a model which describes the changes in cda as a function of age and weight knowing you controlled for sex and height.
Quote Reply
Re: Real time CdA in Zwift? [scott8888] [ In reply to ]
Quote | Reply
scott8888 wrote:
I’m curious about what you are suggesting because I have a professional interest in big data and the issues surrounding bias introduced when doing these sorts of calculations.

Hmmm. It's nice that you have a professional interest in big data. This isn't a big data question. It's pretty small. In this case, what I'm suggesting is change your weight in Zwift and recalculate the implied Crr and CdA.
Quote Reply